Brick by Brick (Volume 2)

4. Tailoring urban policies to the new geography of housing demand

Abstract

The COVID-19 pandemic and the on-going digital transformation of economies and societies have influenced preferences over housing types and locations. People have been spending more time at home, which has created demand for more space and better local amenities. Greater uptake of working-from-home practices has allowed workers to live further away from their place of work. This chapter deals with the evolving geography of housing demand, its drivers and policy implications.

Main policy lessons

With widespread lockdowns and working-from-home mandates, the COVID-19 pandemic has profoundly affected housing markets. In particular, the increased recourse to remote work, at least in sectors of the economy and activities with limited regular need for person-to-person interactions, is likely to have accelerated a long-lasting change in work practices that digitalisation has made possible but might otherwise have taken longer to materialise. These changes are influencing location choices and preferences that have been reshaping housing demand in several OECD countries since the onset of the pandemic, with policy implications that might differ across and within countries depending on local conditions and social preferences.

The main insights for the design of housing policy are:

• In most large cities, house prices decline sharply with the increase in distance to city centres: they exhibit a negative “house price gradient”. These gradients became steeper before the pandemic but have flattened since due to increasing demand for housing in peripheral areas.

• This flattening of the house price gradient has been stronger where the take-up of working from home has been more widespread, corroborating the view that digitalisation is a key driver of the new geography of housing. It has also been stronger where local amenities are better.

• House price increases have been more muted where supply has responded more swiftly to changes in demand.

• Environmental amenities and disamenities strongly influence housing demand.

• Demand for housing has also risen in locations adjacent to metropolitan areas, especially in secondary cities rather than towns or rural areas.

• Priorities for housing policy reform include:

  • Harnessing digital technologies to better match housing demand and supply;

  • Developing digital government solutions and closing the digital skill divide by providing easily accessible lifelong learning and training opportunities;

  • Applying flexible land-use regulations that allow supply to respond to demand within urban strategies incorporating environmental, transport and public-service-delivery objectives;

  • Implementing split-rate housing taxes with higher rates for land than structures to unlock supply and densify urban and suburban areas;

  • Removing obstacles to residential mobility by shifting property taxation away from transaction-based to recurrent taxes;

  • Using land-value capture mechanisms to provide amenities inclusively and fund compensatory measures for low-income households in areas affected by environmental disamenities;

  • Implementing rental-market regulations that, while protecting tenants, include sufficient flexibility to maintain incentives to supply rental housing.

Housing demand varies considerably within metropolitan areas. Following the seminal work by Alonso (1964[1]), Mills (1967[2]) and Muth (1969[3]), a large body of the economic literature has studied the equilibrium between distance to labour markets and residential real estate prices. Jobs and urban amenities are concentrated in the central business district, where space for residential structures is scarce. As workers seek to reduce commuting costs, demand declines with distance to the centre. As a result, house prices and rents generally fall with distance from central business districts in a pattern that is usually called a negative “house price gradient” consistent with the modelling assumption for “monocentric cities”.

The COVID-19 pandemic has affected housing markets by influencing housing preferences and location choices, with implications for the design of housing policy. These changes have been facilitated by digitalisation, which has enabled a rapid increase in remote working, at least in those sectors of the economy and activities with limited need for regular person-to-person interactions. Changing spatial housing demand patterns affect house prices and rents, especially where supply is rigid, with the potential of aggravating affordability challenges in urban spaces. Against this background, this chapter assesses recent trends in the spatial distribution of residential real estate and identifies areas for policy reform to make housing markets operate efficiently and in a manner that addresses affordability and sustainability objectives.

Monitor spatial trends in house prices and residential construction with new data

With more than three years since the pandemic started, there is a growing body of evidence that seems to confirm initial anecdotes of flattening intra-city house price gradients in large metropolitan areas (Figure 4.1).1 The OECD has contributed to this debate by collecting cross-country data on disaggregated house prices (Annex A). The negative house price gradient is stronger for large cities, where commuting costs tend to be higher (Figure 4.2). Major cities are among the functional urban areas (FUA) with the steepest negative gradients, including London, New York, Washington, Mexico City, Paris, Berlin, Hamburg, Brussels, Barcelona and Madrid. However, a few large cities have no significant negative gradient, such as several sprawled metropolitan areas in the United States, Germany’s Ruhr area and England’s West Midlands. Negative house price gradients are seldom observed in smaller and medium-sized FUAs.2

Figure 4.1. House prices decline with distance from the centre in large metropolitan areas

Note: The figure shows average values using local projection (loess filter) across more than 10 000 local area units from more than 500 functional urban areas (FUA) from 16 countries.

Source: OECD Geography of Housing Demand database and OECD calculations.

Statlink

Figure 4.2. House price gradients vary considerably within and across urban areas

Note: House prices for 2018 are demeaned by population-weighted functional urban areas (FUA) average house prices and regressed on the distance to the largest high-density cluster within the FUA (prices and distances in logarithms). The elasticity estimated with this regression is the estimated change in prices for 1% increase in distance from the city centre reported on the vertical axis. Only FUAs with coefficients significant at the 95% level are shown. Colour and size of circles illustrate the population size class of the respective FUA: Small (<200K), Medium (200K-500K), Metro (500K-1.5m), Large (>1.5m).

Source: OECD calculation based on the Geography of Housing Demand database.

Statlink

The evolution of house prices reflects changes in both demand and supply. While supply can be assumed fixed over short horizons, adjustments cannot be ignored over the medium-to-longer term, even considering long construction delays and general scarcity of constructible land in dense urban areas.3 The extent to which supply responds to demand shifts nevertheless varies across countries, as reflected in observable changes in built-up residential areas in OECD countries during 2019-21 (Figure 4.4).

Figure 4.3. The dynamism of residential supply differs across countries

Note: Blue bars denote the median expansion of the residential footprint across each country’s local units. Error bars indicate respectively the 25^th and the 75^th percentile of the increase in residential land use across units.

Source: “Monitoring land use in cities using satellite imagery and deep learning”, (Banquet et al., 2022[4]).

Statlink

Explore the effects of working from home on the spatial distribution of housing demand

The adoption of working from home since the start of the pandemic reduces commuting costs, which influences the intra-city allocation of jobs. Indeed, less frequent commuting makes residential areas far from the urban core more attractive, broadening the range of location choices (Figure 4.4). For instance, a worker who used to commute five times a week for thirty minutes might accept commuting three times a week for fifty minutes. Workers have been working from home to varying degrees across cities and countries (Figure 4.5).

Figure 4.4. 5x30=3x50 minutes: working-from-home widens the potential residential area

Note: Access to economic and social activity proxied by accessibility of 500 restaurants within 30 (50) minutes by car.

Source: OECD Urban Access Framework (OECD/ITF, 2019[5]).

Statlink

Figure 4.5. Working from home has taken off to varying degrees across the world

Note: Working from home proxied by Google Trend subject “Microsoft Teams”. Percentage point changes between early 2020 and early 2022 by region are shown. Green bars denote the median change in working from home across a country’s functional urban areas. Error bars indicate the 25th and the 75th percentiles of that change across regions, respectively. This Google Trend variable is correlated with the share of the total workforce reporting to usually work from home according to the regional module of the 2022 Eurostat Labour Force Survey.

Source: Google Trends and OECD calculations.

Statlink

Urban house price gradients have flattened since the onset of the pandemic, but only in large FUAs (Figure 4.6).4 The shift is stronger for those FUAs where house price gradients were steeper before the pandemic, reflecting higher commuting costs. In zones experiencing strong demand for housing, additional supply is expected to attenuate the pressure on house prices.

Figure 4.6. The urban house price gradient has flattened in large urban areas

Note: Error bars reflect two standard deviations around the estimated coefficient. “Monocentric” functional urban areas (FUA) exhibit a negative price-distance gradient (see Figure 4.2). FUA population size classes are defined as follows: Small (<200K inhabitants), Medium (200K-500K), Metro (500K-1.5m), Large (>1.5m).

Source: OECD calculation based on the Geography of Housing Demand database.

Statlink

Accounting for local residential construction activity thus allows for better identifying changes in housing demand. Indeed, rising house prices could reflect scarce supply, buoyant demand, or both. Disentangling these effects is essential to assess the new geography of housing demand. The distribution of land use within FUAs corroborates stylised facts for population densities as the share of residential land declines with distance to the FUA centre (Figure 4.7, Panel A). But, from 2019 to 2021, satellite images suggest that residential construction activity was more buoyant in peripheral districts of large metropolitan areas (Figure 4.7, Panel B).

Figure 4.7. Residential construction activity increases with distance to the centre

Note: The figures illustrate the distribution of the share of residential land use and its changes across all local area units (districts) as a function of the distance in km to the centre of the functional urban area. The error bars denote the 25^th and 75^th percentiles, respectively.

Source: “Monitoring land use in cities using satellite imagery and deep learning” (Banquet et al., 2022[4]) and OECD calculations.

Statlink

Including these proxies for area-by-area residential construction in the analysis allows for isolating supply effects in the evolution of house prices resulting in a more precise identification of demand pressures.5 The results indicate that accounting for supply effects increases the magnitude and significance of estimated correlations, strengthening the narrative that the observed flattening of urban house price gradients in the wake of the pandemic is related to changing demand patterns (Ziemann et al., 2023[6]).

Furthermore, incorporating the estimated take-up of remote work (Figure 4.5) into the model corroborates indications of a causal link from greater use of remote work, which facilitates living further away from city centres, to flattening house price gradients. Indeed, the flattening was more pronounced in areas with higher take-up of working-from-home practices (continuous dark-blue line in Figure 4.8). By contrast, urban areas with below-average take-up of remote work witnessed hardly any change in the gradient as the change in house prices from 2019 to 2021 barely depended on the distance to the city centre (short-dashed light-blue line in Figure 4.8).

Figure 4.8. More remote work implies weaker price pressure in the centres and stronger pressure in the peripheries of large urban areas

Note: The plots illustrate how the relationship between distance and price changes depends on the level of remote work (specification 4 in Annex B) proxied by the variable gt_teams, which denotes the change in the Google Trend Index topic “Microsoft Teams” from 2019 to 2021: this variable is tightly correlated with a direct measure of working from home while offering the benefit of much wider coverage. For illustrative purposes, three levels of the interacted variables are shown: “Mean” (the average value for remote work, equivalent to the price-distance slope without interaction), “+ 1 SD” (remote work one standard deviation above its average value) and “- 1 SD” (remote work one standard deviation below its average value).

Source: Geography of Housing Database, Google Trends and OECD calculations

Statlink

Price pressures in suburban areas have been weaker, where supply has been more responsive to stronger demand. In contrast, price gradients have flattened less in more densely populated urban areas where construction activity has been less pronounced (Ziemann et al., 2023[6]). Gradients have also flattened less depending on the availability of urban amenities, such as access to open space and the quality of transport infrastructure, which influences location choices. Price gradients have flattened less in lower-valued areas (Figure 4.9) This finding suggests that the move to the suburbs does not occur homogeneously. Local amenities seem to be valued more strongly in remote areas than near the city centre. Indeed, the loss of amenities is typically seen as one of the opportunity costs incurred when moving to the suburbs.

Figure 4.9. Local amenities have a stronger price impact in peripheral areas

Note: The plot illustrates how the relationship between distance and price changes depends on the level of the interacted variable (here: “P_amenity”; see Annex B for details). For illustrative purposes, three levels of the interacted variables are shown: “Mean” (the average value for the respective interacted variable, equivalent to the price-distance slope without interaction), “+ 1 SD” (the value of the interacted variable one standard deviation above its average value) and “- 1 SD” (the value of the interacted variable one standard deviation below its average value). “P_amenity” is a proxy for the value of non-distance-related average characteristics obtained as the residuals from a regression of the logarithm of house price levels on the logarithm of the distance to the FUA centre.

Source: OECD calculations.

Statlink

There is growing evidence that the value of local environmental amenities and disamenities is reflected in house prices. Insofar as higher house prices raise household wealth, the provision of environmental amenities raises the welfare of homeowners. Conversely, would-be owners or renters can face trade-offs between environmental quality and housing affordability. From a welfare perspective, amenity provision is socially desirable as long as the economic value it creates exceeds the costs it entails.

Housing demand has also changed beyond metropolitan areas. When most pandemic-related mobility restrictions were lifted during 2020-21, house prices picked up outside metropolitan centres as well, even beyond their commuting zones [Figure 4.10 and OECD et al. (2021[7])]. Before the pandemic, house price growth was lower in commuting zones and adjacent areas. This pattern suggests that people moving outside the metropolitan boundaries require a certain degree of density that ensures access to key services and amenities. In contrast, rural areas only benefit from the pandemic’s repercussions when located near the urban core.

Figure 4.10. House prices have been rising fast also outside large metropolitan areas: the example of New York City

Source: “The geography of housing demand in times of COVID-19: Are housing markets changing beyond the metropolitan boundaries?”, (Ahrend et al., 2023[8]).

Statlink

Factor in the impact of environmental amenities on housing demand

Much of the amenity value associated with locational choices comes from environmental quality. Open space and access to green spaces played a particular role in location choices during the pandemic. Such amenities provide not only health benefits due to reduced congestion and better air quality, but they also promote social cohesion and improve the quality of life in urban areas, values increasingly demanded since the pandemic outbreak. Places with scarce open space and difficult access to greenery experienced a sharper shift in demand from the city centre to the periphery (Figure 4.11).

Figure 4.11. Environmental quality is a major driver of price shifts away from city centres

Note: The plots illustrate how the relationship between distance and price changes depends on the level of the interacted variable (here, “green_15_walk” and “Open.space”; see Annex B for details). For illustrative purposes, three levels of the interacted variables are shown: “Mean” (the average value for the respective interacted variable, equivalent to the price-distance slope without interaction), “+ 1 SD” (the value of the interacted variable one standard deviation above its average value) and “- 1 SD” (the value of the interacted variable one standard deviation below its average value). “green_15_walk” measures the surface of green space accessible by a 15 minute walk. “Open.space” refers to the local area’s share of open space in land use as inferred from satellite images. “log distance” denotes the demeaned logarithm of the distance to the respective FUA centre.

Source: “Benchmarking Accessibility in Cities - Measuring the Impact of Proximity and Transport Performance” (OECD/ITF, 2019[5]);  “Monitoring land use in cities using satellite imagery and deep learning” (Banquet et al., 2022[4]); Geography of Housing Demand database and OECD calculations.

Statlink

The link between environmental quality and house prices

The positive link between environmental quality and house prices is well established. House prices are affected by environmental amenities, ranging from proximity to open spaces and water bodies, which tend to make residential locations more pleasant, as well as disamenities, such as air and water pollution, noise and proximity to industrial sites and landfills, which have the opposite effect. There is large body of national and international evidence on these effects that provides useful insights for policy.6

Proximity to open spaces and water bodies is associated with higher house prices. This is the case of parks, and to a lesser extent forests and greenbelts (Farrow et al., 2022[9]). Empirical evidence suggests that the positive impact of park views on housing prices ranges between 4% and 8% (Figure 4.12). In addition, proximity to, and views of, water bodies, such as lakes, rivers, streams and oceans, also tend to be associated with higher house prices (Figure 4.13). In some cases, however, there is a negative premium related to proximity to large wetlands, suggesting that they can be seen as both amenities and disamenities depending on contextual and risk factors such as flooding.

Figure 4.12. Effect of a view of open space on house prices

Note: Average impact line is calculated as the average of the results excluding Black and Richards (2020[10]), which is an outlier, reporting an estimate of 35.5%. The outlier could be due to the fact that the amenity in question was the New York High Line, which is a tourist attraction and is arguably more than an environmental amenity. Further, the High Line is located in an area with a particularly high level of real estate development. The source provides bibliographical information on the studies providing the reported estimates.

Source: “Provision of urban environmental amenities: A policy toolkit for inclusiveness”, (Farrow et al., 2022[9]).

Statlink

Figure 4.13. Effect of water body view on house prices

Note: Average impact line is calculated as the average of the results excluding Pompe (2008[11]), which is an outlier due to a particular sample of houses with above-average prices. The source provides bibliographical information on the studies providing the reported estimates

Source: “Provision of urban environmental amenities: A policy toolkit for inclusiveness”, (Farrow et al., 2022[9]).

Statlink

Proximity to industrial infrastructure has mixed, even if by and large negative, effects on house prices. These infrastructures include non-residential sites developed to provide services, such as generating power, disposing of waste or manufacturing products. In the case of power facilities, for example, international experience seems to suggest that wind farms have varied effects on house prices, depending on distance to the facility and the size of turbines (Figure 4.14). A nearby wind farm tends to depress house prices, especially when it is located closer to a residential site and when turbines are taller (Dröes and Koster, 2021[12]). Conventional power plants also have an adverse house price effect, even though this impact subsides with the distance from the site of the plant. Evidence is less clear-cut for nuclear power stations.

Other industrial infrastructures, such as factories, brownfields and landfills, have heterogeneous but generally negative impacts on house prices. In contrast, the clean-up of brownfield sites, where hazardous pollutants may have been disposed of, tends to increase house prices, particularly in areas that had previously been significantly degraded. Landfills have been associated with significant negative impacts on house prices.

Figure 4.14. Decarbonisation efforts can have adverse effects on house prices

Note: The intervals depict the range of estimates in the studies surveyed by the source. The points represent median estimates.

Source: “Provision of urban environmental amenities: A policy toolkit for inclusiveness”, (Farrow et al., 2022[9]).

Statlink

Figure 4.15. Effect on house prices of a one-decibel increase in noise

Note: The size of the blue circles represents the standard deviation of the impact reported in the corresponding article. Articles represented by black dots did not report the standard deviation. The source provides bibliographical information on the studies providing the reported estimates.

Source: “Provision of urban environmental amenities: A policy toolkit for inclusiveness”, (Farrow et al., 2022[9]).

Statlink

Air and noise pollution tend to depress house prices. The negative price premium reflects the umpleasantness associated with a degraded environment. Homes in areas with poor water quality also tend to be less sought after than those located in areas with better quality (Young, 1984[13]; Gibbs et al., 2002[14]).

Contextual factors may dampen or enhance the impacts of environmental amenities on house prices. For instance, greater awareness of the presence of local parks and the benefits of green spaces to residents tends to increase the impact of such amenities on local house prices. By the same token, information on flood risk and awareness of the risks of a local river also influence house prices (Chen, Li and Hua, 2019[15]; OECD, 2018[16]).

Population density affects the benefits associated with urban environmental amenities. Higher density tends to be associated with greater willingness to pay for open space, while fostering awareness about the benefits of amenities, which can enhance their impact on house prices. On the other hand, density may reduce the benefits of environmental amenities to the extent that crowding makes an amenity less enjoyable (Neuts, Nijkamp and Leeuwen, 2012[17]).

Willingness to pay for an amenity is only a partial measure of the total benefits the amenity generates for society. For example, improved long-term health outcomes resulting from better air quality are a benefit of urban forest creation that is not necessarily reflected in house prices. If environmental amenities create indirect costs or benefits, they create a wedge between house prices and the true social value of the amenities. Changes in house prices associated with a change in environmental amenity provision can therefore be thought of as only a partial reflection of potential welfare changes.

Behavioural biases and incomplete information may also cause willingness to pay to be an incomplete welfare measure. Home buyers often have at least some missing information regarding purchased properties. Other biases, for example, overvaluation fuelled by social contagion or disproportional aversion to losses compared to equivalent gains, can contribute to inaccurate individual valuations of environmental amenities (Salzman and Zwinkels, 2017[18]). These inaccuracies in valuations can eventually be reflected in house prices.

Distributional impacts of amenity provision

Access to environmental amenities is unequally distributed among social groups. Disadvantaged populations are disproportionately exposed to disamenities, such as pollution, poor water quality, and proximity to hazardous waste facilities and brownfield sites (Farrow et al., 2022[9]), OECD (2006[19]). Inequalities in the initial distribution of environmental disamenities imply that policies to improve environmental quality will be progressive insofar as those closest to disamenities experience the greatest benefits from their remediation (Banzhaf, Ma and Timmins, 2019[20]). As a result, the clean-up of brownfield sites can deliver disproportionate benefits across groups. Whether environmental amenity provision exacerbates or reduces inequality depends on the factors driving any existing inequities in amenity or disamenity provision across socio-economic status, as well as any effects beyond house price changes that their provision may entail (e.g., changes in transport times or quality of life).

Homeownership also plays a vital role in the impact of local amenity provisions on housing affordability. Rises in house prices resulting from amenity provision may decrease affordability for renters. Unlike homeowners, renters do not receive capital gains from housing price increases. They may also be more vulnerable to increases in housing costs, as they typically spend a larger part of their budget on housing-related expenses. Among renters, low-income households have a higher housing cost burden, compared with higher-income households.

Households that use a large share of their income to cover housing-related costs are particularly exposed to the distributional impacts of amenity provision. The displacement of renters and low-income households due to environmental amenity provision can occur via two mechanisms.

• First, as amenity provision in an area increases, the area attracts higher-income households that can pay more for the locally provided amenity. For instance, converting empty space into an urban park will attract potential buyers and renters who value public open space and related recreational facilities. Renters with a low amenity valuation may then be effectively forced to relocate.

• Second, inelastic housing supply, i.e., housing supply that is relatively unresponsive to rising prices, exacerbates the displacement of low-income renters over time in areas with weak rent control. Preventing exclusionary housing patterns near environmental amenities entails accompanying environmental amenity provision with measures that address supply and demand-side aspects. Accordingly, policy measures would do well to maintain an adequate supply of affordable housing in amenity-rich areas and support low-income households that face rent increases due to amenity-induced increases in house values.

Monetary, financial and tax policies determine the relative access to loans and the borrowing cost at which renters can become owners. Falling interest rates temporarily facilitate home ownership, allowing buyers who enter the market at the right time to collect the social value of amenity provision that capitalises into property prices. However, low interest rates also increase house prices by boosting the demand for housing. Mortgage interest relief also reduces the cost of borrowing, temporarily facilitating house purchases before the worth of the relief becomes capitalised into house prices (Chapter 3). Moreover, changes in demand and property values will not be uniform over space, with the provision of local amenities and building-rights scarcity playing a central role in the asymmetric distribution of capital gains.

Adapt policies

The changing patterns in housing demand arising from digitalisation require policy action to unlock supply and avoid the build-up of new inequalities. Policymakers can consider several options depending on local conditions, social preferences and policy settings.

Digital transformation

Digitalisation offers several options for technological change and innovation in the construction and “smart” management of buildings, not least through artificial intelligence and the internet of things. Innovations in urban planning and management are already taking place and can improve the management of traffic, urban amenities and infrastructure, as well as the energy efficiency of buildings and cities at large. Digital platforms can enhance competition and improve the matching of supply and demand for dwellings. Housing fintech can broaden access to finance and reduce borrowing costs to the extent that these activities are regulated appropriately to avoid new sources of financial risk (Chapter 3).

Effective digital infrastructure is a prerequisite for digitisation and remote work. Tackling the digital divide between urban and rural areas is particularly important in the context of spatial shifts in the demand for housing. Governments would do well to ensure widespread access to high-speed internet, upgrade technical and managerial skills, and implement product and labour market reforms to facilitate the uptake of digital technologies by firms (Sorbe et al., 2019[21]). The “OECD Going Digital” project aims to help policymakers better understand the digital transformation and develop appropriate policies to help shape a positive digital future (OECD, 2020[22]).7

Urban planning

Land-use regulations are important policy tools to respond to the new geography of housing demand within urban strategies compatible with environmental, transport and public service provision objectives. Shifts in the geography of housing demand magnify the benefits of regularly revising geographic boundaries for urban development to accommodate city growth while ensuring forms of expansion compatible with environmental objectives. Where regulations allow it, flexibility to convert commercial property and office space for residential use would facilitate the reallocation of housing capital to evolving demand for different uses, potentially making housing more affordable. However, there is a risk that disaffection for city centres gives rise to housing segregation as the better-off move away. These trends pose challenges for urban planning, land-use design and zoning regulations.

The governance of urban planning is often fragmented across government levels and sometimes across ministries or government agencies (OECD, 2021[23]). This situation can complicate reforms if public bodies with responsibility over one area, for instance, land regulation, do not have authority in other areas, such as taxation or social housing, that would allow them to design integrated reform packages. Responsibilities and decision-making should be delegated to the metropolitan rather than the local level to avoid not-in-my-backyard dynamics and foster inter-municipal co-operation, including in the provision of public services and transport. In other cases, there is a merit in enhancing tax and spending autonomy at the local level to boost housing supply responsiveness, especially where policymaking functions are overly centralised at higher levels of administration (Dougherty, Cournède and van Hoenselaar, 2023[24]).

The new geography of housing demand entails increasing distances between workplaces and residences. Fewer but also longer commutes might require rethinking urban passenger transport systems, notably in light of continuing efforts to decarbonise transport. Compact and transit-oriented development, commonly defined as mixed-use urban development with mass-transit facilities within walking distance of residential buildings, can make public transport more convenient, encourage ridership and decrease car dependency (ITF, 2019[25]).

Amenity provision

Environmental policies should support the provision of environmental amenities when net welfare benefits exceed net costs. Despite their myriad benefits, environmental amenities remain undersupplied in many urban areas. Public investments with strong spatial dimensions, such as the provision of open space, can increase house prices and burden residents who do not benefit from the associated gains, leading to an unequal distribution of the net benefits of the provided goods. Such effects should be anticipated by environmental policies to support amenity provision, and particular attention should be paid to the distributional effects that occur between renters and owners in affected areas. Additional attention should be given to displacement effects, whereby residents can face pressure over time to relocate out of the area due to amenity-induced increases in housing costs.

Amenity provision may need to be accompanied by complementary measures designed to mitigate economy-wide effects (OECD, 2018[26]) and negative distributional impacts that can occur via housing markets. Examples of policy measures that address both environmental issues and equity include subsidising the retrofitting of the existing housing stock (Chapter 2) and investments in green social housing, i.e., social housing that incorporates environmental amenities (OECD, 2021[23]). Enabling portable eligibility with respect to social housing could also be included in the toolkit of feasible interventions (OECD, 2021[27]). Relaxing building height restrictions can also improve access to environmental amenities.

Low-income areas tend to be overlooked in green renewal project planning (Haase et al., 2017[28]; Anguelovski et al., 2016[29]). There is scope to introduce greater equity into the urban planning process at the earliest stages of such projects. One means of doing so is to facilitate the inclusion of residents of all socio-economic status in participatory planning processes. Enhancing amenity provision in amenity-scarce areas would generate greater marginal benefits than their provision in areas with substantial existing supply. Policymakers should aim for a more uniform distribution of amenities, which may involve targeting areas with little supply.

Existing local fiscal and land-use policies, public finance mechanisms, as well as the spatial profile of amenities, tenure status and income, determine the potential for distributional impacts. As a result, these conditions should be taken into account when evaluating the appropriateness of potential compensation mechanisms. For example, successful implementation of land value capture mechanisms should take into account factors such as the maturity of land markets, land use regulations, investment policies, legal frameworks, fiscal and governance structures, as well as local circumstances and conventions regarding land rights (OECD, 2021[27]). The welfare impact of property taxes, for example, will depend not only on their magnitude but also on the relation between the tax rate on land and differences in rates across different land use categories (Brandt, 2014[30]).

Land value capture

Land value capture measures, including infrastructure levies or developer obligations, as well as smart ways to manage and re-adjust land use, can incentivise development and help densify existing residential areas (OECD/Lincoln Institute of Land Policy, PKU-Lincoln Institute Center, 2022[31]). Such measures can also contribute to financing the infrastructure and amenities needed to improve the accessibility of economic and social facilities in remote areas. Numerous land value capture strategies exist. These include betterment contributions and special assessments, impact fees, land readjustment, and inclusionary zoning (OECD, 2021[27]; Farrow et al., 2022[9]):

• Betterment contributions and special assessments require owners of properties benefitting from a public investment to pay the municipality a fee. The fee is assessed based on the added property value that the owners’ profit from due to the public investment.

• Impact fees are similar to betterment contributions and special assessments, except that they are charged in the form of a one-time fee.

• Land readjustment occurs when landowners collaborate with a municipality to pool land that will be devoted to amenity development. Following the development of the pooled land, each landowner receives a smaller parcel that has greater value due to the provision of the created amenity.

• Inclusionary zoning involves setting minimum thresholds for the proportion of low- or moderate-income housing that developers should provide in exchange for the right to construct residential properties, and could be used in conjunction with amenity provision to ensure affordable housing supply in areas close to an environmental amenity.

Potential uses of these revenues vary according to the specific context and can include investment in social housing and the provision of housing subsidies for low-income households. Pricing mechanisms can also be used to recover the costs generated by environmental disamenities. The revenues generated by these pricing mechanisms can serve to compensate households that may disproportionately suffer from the impacts of disamenities.

Housing taxation

The design of housing taxes influences residential mobility. The use of transaction taxes is generally assessed as generating efficiency losses mainly through lock-in effects that hold back residential mobility (OECD, 2022[32]). In addition, reliance on transaction taxes may strengthen incentives to buy less expensive land, which generally lies far from city centres and transport infrastructure, while deterring transactions that might help put land to more efficient uses, including residential ones. They also encourage the purchase of undeveloped land for new development rather than upgrading developed areas (Blöchliger and Kim, 2016[33]). Substituting at least partly recurrent property taxes for transaction taxes would make tax systems and housing markets more efficient with benefits for residential and labour mobility (OECD, 2022[32]).

Despite their efficiency, there is significant scope to improve the design and functioning of recurrent taxes on immovable property (OECD, 2022[32]). While all OECD countries levy recurrent taxes on immovable property, they are in many instances based on outdated property values, significantly reducing the associated revenue potential, while harming equity and economic efficiency. Opting for a split-rate design, whereby land is taxed at a higher rate than structures, would encourage the development of vacant or underused land in suburban areas, thereby fostering compact development and attenuating urban sprawl.

Property taxes provide local governments with stable revenue to finance the provision of local public goods and services, which in turn is a key determinant of residential settlement decisions, particularly for residents planning to relocate from amenity- and service-rich urban centres. In areas where housing supply shortages coincide with an abundance of vacant homes, recurrent taxes on these vacant homes can help increase housing supply. Such taxes should be flanked by credible measures to monitor compliance and avoid loopholes for short-term rentals (OECD, 2022[32]).

Rental market policies

Rental-market regulation can hinder supply when they involve overly tight rent controls (OECD, 2021[23]). Strict tenant-landlord regulation resulting in high tenure security and rent control can lower the expected returns from the residential rental supply, thereby reducing residential investment or encouraging alternative uses of the existing stock by owners.

Tight rental contract restrictions also adversely affect vulnerable renters, posing obstacles to residential and labour mobility. Excessive protection of tenants implies that renters with uncertain labour market prospects find it difficult to sign a lease. Strict regulations in rental markets further reduce residential mobility, as tenants in rent-controlled dwellings will be reluctant to move if rents are below market levels. As there is also a case for providing tenants with reasonable security over tenure and rent levels, a balanced system can involve a degree of rent stabilisation, whereby rents can be adjusted for new contracts (and potentially renewals) but regulated in line with market developments during the duration of a contract.

References

[8] Ahrend, R. et al. (2023), Expanding the doughnut: the impact of remote work and COVID-19 on the geography of housing demand, OECD Publishing.

[34] Ahrend, R. et al. (2022), “Changes in the geography housing demand after the onset of COVID-19: First results from large metropolitan areas in 13 OECD countries”, OECD Economics Department Working Papers, No. 1713, OECD Publishing, Paris, https://doi.org/10.1787/9a99131f-en.

[1] Alonso, W. (1964), Location and land use. Toward a general theory of land rent., Cambridge, Mass.: Harvard Univ. Pr., https://www.cabdirect.org/cabdirect/abstract/19641802976.

[29] Anguelovski, I. et al. (2016), “Equity Impacts of Urban Land Use Planning for Climate Adaptation: Critical Perspectives from the Global North and South”, https://doi.org/10.1177/0739456X16645166, Vol. 36/3, pp. 333-348, https://doi.org/10.1177/0739456X16645166.

[4] Banquet, A. et al. (2022), “Monitoring land use in cities using satellite imagery and deep learning”, OECD Regional Development Papers, No. 28, https://www.oecd-ilibrary.org/urban-rural-and-regional-development/monitoring-land-use-in-cities-using-satellite-imagery-and-deep-learning_dc8e85d5-en (accessed on 8 November 2022).

[20] Banzhaf, S., L. Ma and C. Timmins (2019), “Environmental Justice: The Economics of Race, Place, and Pollution”, Journal of Economic Perspectives, Vol. 33/1, pp. 185-208, https://doi.org/10.1257/jep.33.1.185.

[10] Black, K. and M. Richards (2020), “Eco-gentrification and who benefits from urban green amenities: NYC’s high Line”, Landscape and Urban Planning, Vol. 204, p. 103900, https://doi.org/10.1016/J.LANDURBPLAN.2020.103900.

[33] Blöchliger, H. and J. Kim (eds.) (2016), Fiscal Federalism 2016: Making Decentralisation Work, OECD Publishing, Paris, https://doi.org/10.1787/9789264254053-en.

[30] Brandt, N. (2014), “Greening the Property Tax”, OECD Working Papers on Fiscal Federalism, No. 17, OECD Publishing, Paris, https://doi.org/10.1787/5jz5pzw9mwzn-en.

[15] Chen, W., X. Li and J. Hua (2019), “Environmental amenities of urban rivers and residential property values: A global meta-analysis”, Science of The Total Environment, Vol. 693, p. 133628, https://doi.org/10.1016/J.SCITOTENV.2019.133628.

[24] Dougherty, S., B. Cournède and F. van Hoenselaar (2023), Housing supply responsiveness across levels of government: tax & spending autonomy, OECD Publishing.

[12] Dröes, M. and H. Koster (2021), “Wind turbines, solar farms, and house prices”, Energy Policy, Vol. 155, p. 112327, https://doi.org/10.1016/J.ENPOL.2021.112327.

[9] Farrow, K. et al. (2022), “Provision of urban environmental amenities: A policy toolkit for inclusiveness”, OECD Environment Working Papers, No. 204, OECD Publishing, Paris, https://doi.org/10.1787/0866d566-en.

[14] Gibbs, J. et al. (2002), “An hedonic analysis of the effects of lake water clarity on New Hampshire lakefront properties”, cambridge.org, https://www.cambridge.org/core/journals/agricultural-and-resource-economics-review/article/an-hedonic-analysis-of-the-effects-of-lake-water-clarity-on-new-hampshire-lakefront-properties/7D3AE434A407BF9DCB11F289DDC4F57B (accessed on 15 October 2021).

[35] Gokan, T. et al. (2022), “How the Rise of Teleworking Will Reshape Labor Markets and Cities”, SSRN Electronic Journal, https://doi.org/10.2139/SSRN.4235466.

[36] Gupta, A. et al. (2022), “Flattening the curve: Pandemic-induced revaluation of urban real estate”, Journal of Financial Economics, Vol. 146/2, pp. 594-636, https://doi.org/10.1016/J.JFINECO.2021.10.008.

[28] Haase, D. et al. (2017), “Greening cities – To be socially inclusive? About the alleged paradox of society and ecology in cities”, Habitat International, Vol. 64, pp. 41-48, https://doi.org/10.1016/J.HABITATINT.2017.04.005.

[37] Huang, N., J. Pang and Y. Yang (2022), “COVID-19 and household preference for urban density in China”, Journal of Urban Economics, p. 103487, https://doi.org/10.1016/j.jue.2022.103487.

[25] ITF (2019), ITF Transport Outlook 2019, OECD Publishing, Paris, https://doi.org/10.1787/transp_outlook-en-2019-en.

[2] Mills, E. (1967), “An Aggregative Model of Resource Allocation in a Metropolitan Area”, The American Economic Review, Vol. 57/2, pp. 197-210, https://doi.org/10.2307/2981088.

[3] Muth, R. (1969), Cities and Housing. The Spatial Pattern of Urban Residential Land Use, Cambridge University Press, https://doi.org/10.1017/s0770451800027500.

[17] Neuts, B., P. Nijkamp and E. Leeuwen (2012), “Crowding Externalities from Tourist Use of Urban Space:”, http://dx.doi.org/10.5367/te.2012.0130, Vol. 18/3, pp. 649-670, https://doi.org/10.5367/TE.2012.0130.

[32] OECD (2022), Housing Taxation in OECD Countries, OECD Tax Policy Studies, No. 29, OECD Publishing, Paris, https://doi.org/10.1787/03dfe007-en.

[23] OECD (2021), Brick by Brick: Building Better Housing Policies, OECD Publishing, Paris, https://doi.org/10.1787/b453b043-en.

[27] OECD (2021), Building a Global Compendium on Land Value Capture - OECD, https://www.oecd.org/cfe/cities/Land-Value-Capture.htm (accessed on 3 November 2021).

[22] OECD (2020), “Going Digital Integrated Policy Framework”, https://doi.org/10.1787/dc930adc-en.

[16] OECD (2018), Cost-Benefit Analysis and the Environment: Further Developments and Policy Use, OECD Publishing, Paris, https://doi.org/10.1787/9789264085169-en.

[26] OECD (2018), “Steering urban development to more sustainable pathways”, in Rethinking Urban Sprawl: Moving Towards Sustainable Cities, OECD Publishing, Paris, https://doi.org/10.1787/9789264189881-7-en.

[19] OECD (2006), The Distributional Effects of Environmental Policy, Organisation for Economic Co-operation and Development, Paris, https://www.oecd-ilibrary.org/environment/the-distributional-effects-of-environmental-policy_9789264066137-en.

[5] OECD/ITF (2019), Benchmarking Accessibility in Cities - Measuring the Impact of Proximity and Transport Performance, https://www.itf-oecd.org/benchmarking-accessibility-cities.

[31] OECD/Lincoln Institute of Land Policy, PKU-Lincoln Institute Center (2022), Global Compendium of Land Value Capture Policies, OECD Regional Development Studies, OECD Publishing, Paris, https://doi.org/10.1787/4f9559ee-en.

[7] OECD et al. (2021), Applying the Degree of Urbanisation: A Methodological Manual to Define Cities, Towns and Rural Areas for International Comparisons, OECD Regional Development Studies, OECD Publishing, Paris/European Union, Brussels, https://doi.org/10.1787/4bc1c502-en.

[11] Pompe, J. (2008), “The effect of a gated community on property and beach amenity valuation”, Land Economics, Vol. 84/3, pp. 423-433, http://le.uwpress.org/content/84/3/423.short (accessed on 20 October 2021).

[38] Reusens, P., F. Vastmans and S. Damen (2022), “The impact of changes in dwelling characteristics and housing preferences on house price indices”, NBB Working Paper Series, No. 406, National Bank of Belgium, https://www.nbb.be/en/articles/impact-changes-dwelling-characteristics-and-housing-preferences-house-price-indices (accessed on 15 February 2023).

[18] Salzman, D. and R. Zwinkels (2017), “Behavioral Real Estate”, Journal of Real Estate, Vol. 25/1, pp. 77-106, https://www.jstor.org/stable/26391901 (accessed on 13 December 2021).

[21] Sorbe, S. et al. (2019), “Digital Dividend: Policies to Harness the Productivity Potential of Digital Technologies”, OECD Economic Policy Papers, No. 26, https://doi.org/10.1787/273176bc-en.

[13] Young, C. (1984), The Influence of Water Quality on the Value of Recreational Properties Adjacent to St. Albans Bay, Vermont, https://books.google.com/books?hl=en&lr=lang_en&id=_cYgktwsvBMC&oi=fnd&pg=PR5&dq=The+Influence+of+Water+Quality+on+the+Value+of+Recreational+Properties+Adjacent+to+St.+Albans+Bay,+Vermont&ots=cC2ZSnsFUc&sig=DtLAZA2uy-xTLB0k8-ozxItPBmo (accessed on 15 October 2021).

[6] Ziemann, V. et al. (2023), “Tailoring urban policies to the new geography of housing demand”, OECD Economics Department Working Papers, OECD Publishing, Paris, forthcoming.

Annex 4.A.  The OECD Geography of Housing Demand Database

The OECD, supported by a network of public and private data providers, has assembled the Geography of Housing Demand (GHD) database of housing transaction prices for 16 countries at the smallest administrative unit available (Ziemann et al., 2023[6]; Ahrend et al., 2022[34])). In most cases, the data are made available by national statistical agencies that collect them to compile HPIs. France and the United Kingdom publish open-source data for every single transaction. Finally, private data operators from Germany (“vdpResearch”) and Portugal (“Confidencial Imobiliário”) agreed to share granular house price data computed from their proprietary databases.

Annex Table 4.A.1. Data sources and coverage by country

	Coverage			Number of FUAs (districts)				Source	Local house price variable
	Period	Population (%)	Area (%)	Small (<200K)	Medium (200K – 500K)	Metro (500K – 1.5m)	Large (> 1.5m)
AUT	2017Q1 - 2022Q1	62	24	-	3 (67)	2 (70)	1 (123)	Statistik Austria	Median transacted price per m2
BEL	2017Q1 - 2021Q4	94	77	5 (16)	4 (35)	4 (105)	1 (125)	STATBEL	Average transacted price per m2
DEU	2018Q1 - 2021Q4	74	60	12 (76)	54 (751)	17 (698)	8 (1293)	vdpResearch	Average transacted price per m2
DNK	2017Q4 - 2022Q1	46	25	-	3 (67)	-	1 (68)	Statistics Denmark	Average transacted price per m2
ESP	2017Q1 - 2021Q3	92	51	46 (277)	22 (225)	8 (222)	3 (383)	INE	Average transacted price per m2
FIN	2017Q1 - 2021Q4	81	37	3 (15)	3 (31)	1 (15)	-	Statistics Finland	Average transacted price per m2
FRA	2017Q1 - 2021Q4	86	66	17 (1075)	30 (3301)	13 (2338)	2 (2071)	Demande de valeurs foncieres	Median transacted price per m2
GBR	2017Q1 - 2022Q3	86	62	36 (570)	27 (1233)	14 (1845)	4 (2556)	UK Government Price Paid data	Average of median transacted prices per type and age of property
HUN	2017Q1 - 2022Q1	92	74	11 (263)	7 (304)	-	1 (201)	Hungarian Central Statistics Office	Average transacted price per m2
ISR	2017Q1 - 2021Q4	65	12	1 (5)	4 (27)	2 (25)	1 (59)	Central Bureau of Statistics	Average transacted price per m2
KOR	2018Q1 - 2021Q4	81	26	1 (1)	10 (11)	6 (22)	5 (105)	MOLIT	Average transacted price per m2
MEX	2017Q1 - 2021Q4	54	2	18 (246)	36 (885)	30 (1847)	8 (2070)	Sociedad Hipotecária Federal (SHF)	Average transacted price per m2
NOR	2017Q1 - 2022Q1	65	12	2 (2)	3 (9)	1 (11)	-	Statistics Norway	Average transacted price per m2
PRT	2017Q1 - 2021Q4	44	8	5 (15)	2 (22)	1 (43)	1 (110)	Confidencial Imobiliário	Average transacted price per m2
SWE	2017Q1 - 2021Q4	90	98	6 (22)	3 (13)	2 (18)	1 (21)	Svensk Mäklarstatistik	Average transacted price per m2
USA	2017Q1 - 2022Q3	96	56	34 (496)	84 (2181)	59 (3263)	34 (6617)	Zillow Research Institute	Zillow Home Value Index

Annex 4.B. Measuring the urban house price gradient and its changes

The following specification assesses whether house prices vary with distance to the city centre:

$\ln P_{i,j}={\alpha }_j+{\beta }_j\ln D_i^j+{\epsilon }_{i,j}$ (1)

where $P_{i,j}$denotes the house price in local unit i of FUA j in 2018 and$D_i^j$ the distance in metres from local area i to the centroid of the largest high-density cluster of the corresponding FUA j. The estimated coefficient ${\beta }_j$ is referred to as gradient.

The core hypothesis tested is whether intra-FUA house price gradients have flattened in the wake of the COVID pandemic as remote work practices have become more widespread. To do so, the change in local house prices between the second half of 2019 (pre-COVID) and the second half of 2021 (the latest uniformly available data since the COVID outbreak) is regressed on the distance to the corresponding FUA centroid.

$\mathrm{\Delta }{P}_{i,j}=\mathrm{\alpha }+\mathrm{\delta }\ln D_i^j+{\mathrm{\mu }}_{\mathrm{j}}+{\epsilon }_{i,j}\mathrm{ }$ (2)

with $\mathrm{\Delta }$ a log-difference operator (measuring the from 2019H2 to 2021H2 in per cent) and ${\mathrm{\mu }}_{\mathrm{j}}$ FUA-fixed effects. A positive slope coefficient $\mathrm{\delta }$ implies a flattening of the intra-FUA house price gradient ${\beta }_j$ from the level equation (1).

To better identify demand shocks in price movements, the reduced form baseline estimation (equation 2) is augmented by a proxy for new residential construction (S_i):

$\mathrm{\Delta }{P}_{i,j}=\mathrm{\alpha }+{\mathrm{\delta }}_0\ln D_i^j+{\mathit{\delta }}_{\mathit{S}}\left(\mathbf{ln}{\mathit{D}}_{\mathit{i}}^{\mathit{j}}\times \mathit{ }{\mathit{S}}_{\mathit{i}}\right)\mathit{ }{+\mathrm{ }\mathbf{\gamma }\mathbf{ }{\mathit{S}}_{\mathit{i}}+\mathrm{\mu }}_{\mathrm{j}}+{\epsilon }_{i,j}\mathrm{ }$ (3)

$\mathrm{\Delta }{P}_{i,j}=\mathrm{\alpha }+{\mathrm{\delta }}_0\ln D_i^j+{\mathbf{\delta }}_{\mathbf{W}\mathbf{F}\mathbf{H}}(\mathbf{ln}{\mathit{D}}_{\mathit{i}}^{\mathit{j}}\times \mathit{ }\mathit{W}\mathit{F}{\mathit{H}}_{\mathit{j}})+{\delta }_S\left(\ln D_i^j\times S_i\right) {+\mathrm{ }\mathrm{\gamma }\mathrm{ }{S}_i+\mathrm{\mu }}_{\mathrm{j}}+{\epsilon }_{i,j}$ (4)

Accordingly, the coefficient $\mathrm{\delta }$ can be rewritten as a function of the rise in working from home (WFH) and additional supply (S) weighted by the respective elasticities:

${\delta }_{i,j}={\delta }_0+{\delta }_{WFH}WF{H}_j+{\delta }_s{S}_i$

• Hypothesis 1: $\mathrm{\delta }$ and ${\mathrm{\delta }}_0$ are positive, reflecting a flattening of the house price gradient since the COVID outbreak.

• Hypothesis 2: ${\delta }_{WFH}$ is positive in line with the assumption that a higher take-up of WFH increases, all other things equal, demand for more remote areas.

• Hypothesis 3: ${\delta }_s$ is negative since, all other things equal, more supply relieves demand pressure on prices.

Notes

← 1.  Gupta et al. (2022[36]) show that house and rent prices decline with increasing distance to the city centre in most US metropolitan areas. The extent to which gradients decline depends on the intensity of working from home and the supply responsiveness in cities to accommodate changing housing preferences. Huang, Pang and Yang (2022[37]) show that the onset of COVID has reduced the gradient in Chinese cities as preferences have shifted towards low-density areas associated with lower infection risks. Gokan et al. (2022[35]) found a significant reduction in the house price gradient in the London area. See Ziemann et al (2023[6]) for additional bibliographic references.

← 2.  The house price gradient could be underestimated because transacted house prices do not take into account the lower average quality of city houses. While the analysis controls for difference in the average size of the houses by using square meter prices, it does not correct for the older age, possible lower energy performance and the larger prevalence of terraced houses in cities (Reusens, Vastmans and Damen, 2022[38]).

← 3.  New geospatial data sources and state-of-the-art machine-learning algorithms open the door to tracking construction activity in almost real time. Recent OECD work has trained an image segmentation model on Sentinel satellite imagery data using the Copernicus Urban Atlas to identify and track different forms of land use, notably including “residential”, “commercial and industrial”, “transport infrastructure”, “open space” and “water and wetlands” (Banquet et al., 2022[4]).

← 4.  The econometric assessment of the impact of working-from-home practices on urban house price gradients builds on earlier explorations by the OECD (Ahrend et al., 2022[34]). The novel dataset includes house transaction prices and transaction volumes from more than 600 FUAs across 16 countries. Annex B describes the methodology to investigate a flattening of urban house price gradients. The results are presented by sub-sample according to the FUAs’ size and the degree to which it is consistent with the monocentric model: FUAs with a significantly negative house price gradient are labelled “monocentric FUAs”.

← 5.  See Specification 3 in Annex B.

← 6.  In order to better compare the results obtained across studies, which can use different metrics, reported results are harmonised where possible to marginal effects in terms of the percentage change in housing prices. For example, For studies assessing the impact of distance to open space, monetary values were converted to percentage changes where possible by using the average house price in the study sample. Nevertheless, direct comparisons remain impossible for some studies, the results of which are therefore generally excluded from the figures in this chapter.

← 7.  The “Going Digital Toolkit” provides a roadmap to policymakers by identifying policies and regulations to reduce connectivity divides. Such policies include promoting competition, fostering investment, and removing barriers to broadband deployment, as well as a set of approaches to extend connectivity in rural and remote areas. The development of digital government services can ensure a better and more inclusive response to citizens’ needs and improve access to public services in disadvantaged communities. Closing the digital skills divide by providing all citizens with ICT, literacy and numeracy skills requires equal opportunities for training, education, re-skilling and upskilling for the jobs and societies of the future.